This article discusses the meaning of reading literary texts as data based on the case of a lecture at a semester that explored the research method of digital humanities using Lee In-jik’s “new novel (sinsoseol)” as a target of analysis. Because linguistic processing of literary texts have a narrative characteristic, it is difficult to apply the same method being applied to non-literary texts to a literary text analysis. Linguistic processing of literature, especially the text of novel, needs to be made not in terms of words or concepts, but at least at the level of verses and between sentences and paragraphs. In this article, I propose to analyze the novel text by segmenting it into semantic units, just as Roland Barthes segmented ‘Sarrasine’ into 561 reading comprehension units in S/Z. Analysis based on semantic units can be divided into metalinguistic data representation and semantic performance data representation. The former conceptualizes semantic units or categorizes them under a specific classification system, contributing to the simplicity and abstract expression of data. However, it is difficult to reflect the chain process of meaning, so it is necessary to follow the plot of the novel and consider the semantic unit as an actual analysis object. That is a semantic performance-based data representation method. The semantic unit of the text is determined by the reader of the text. The extracted semantic unit is further cut into detailed elements that make up the meaning. It is a matter of reframing the network of relationships that form meaning. Because the network of semantic units is expressed as a data set, and the flow of this data set follows the plot of the novel, it shows the different narrative aspects of the novel. Therefore, extracting the semantic unit is also a task that unfolds the researcher’s reading comprehension of the text. In that sense, a semantic performance data representation is a “scriptable” text that Barthes spoke. Segmenting the text and restructuring the relational network by scattering the segmented units are a non-linear reading of the text. Here, the text loses both totality and externality. The image of these texts resembles that of the network. The network has no specific entrance or exit, and the divided units have a fragmented order. However, formality, blockade, clustering, and closure are also important characteristics of networks. In addition, narrative-based networks have temporality. A network with strong nonlinearity is passive in containing temporality. but I suggest that a little more discussion on the model of the network that reflects the narrative is needed for literary research.
